The ability to target gene and drug delivery to defined and limited tissues is one of the key hurdles for effective therapies. This application builds on the system developed in the laboratory that has successfully screened for productive retroviral entry into targeted cells through selection of novel Env/receptor conjugate pairs. Through the use of random libraries generated within the retroviral Env receptor-binding domain, productive infection using alternative host-cell receptors is rapidly obtained. Two major benefits of this system is that the random library is generated within the context of the Env backbone, guaranteeing that the presentation of the novel receptor binding region is in the conformation required for viral entry and that the target cell can essentially be uncharacterized. The experiments in this proposal move this system beyond the basic proof-of- principal, to address the next critical set of milestones. Questions include how many unique, high titer isolates can be obtained from a single cell line? What classes of receptors are selected for productive infection? Can receptor selection be biased towards specific cell surface proteins? Are these vectors suitable for delivery in animal models and what is the most effective means of delivery? Can these novel Env isolates be incorporated into vectors capable of infecting quiescent cells? These questions are addressed focusing on human osteosarcoma (OS) cells as the model system. Two novel Env isolates (L1 and CP) with distinct specificity towards OS cells have been isolated. Experiments aim at defining the range as well as the limitations of the random library screen through characterizing a panel of OS targeting Env isolates with respect tissue specificity, viral interference groups and receptor usage. Protocols to bias the receptor selection through overexpression of the putative receptor protein on the target cell are tested. Experiments that move the test system from in vitro tissue culture systems to in vivo tumor animal models are developed. To date, the known receptors for Env proteins within the MuLV and FeLV Env backbones have utilized multiple membrane-pass proteins. To broaden the class of potential receptors to include single-membrane pass protein receptors, as well as to extend the system to include non-dividing cells, the random Env library will be incorporated into a modified Sindbis Env protein pseudotyped onto lentiviral particles. Collectively, these experiments improve the technology with the goal of identifying novel viral Env vectors targeting entry to specific cells for a broad range of gene delivery protocols. Project Narrative: The safety and effectiveness of retroviral gene delivery would be greatly enhanced with the ability to direct viral infection to limited and targeted cells. This application develops methods to selectively screen for retroviral entry into targeted cells, using human osteosarcoma cells as the model system. The development of retroviral vectors capable of specific entry into targeted cell types is key for the clinical use of retroviral vectors for gene delivery and cancer therapies.